Openstack Cloud Driver



OpenStack is an open source project that is in use by a number a cloud providers, each of which have their own ways of using it.

This OpenStack driver uses a the shade python module which is managed by the OpenStack Infra team. This module is written to handle all the different versions of different OpenStack tools for salt, so most commands are just passed over to the module to handle everything.


There are two ways to configure providers for this driver. The first one is to just let shade handle everything, and configure using os-client-config and setting up /etc/openstack/clouds.yml.

    region_name: RegionOne
      username: 'demo'
      password: secret
      project_name: 'demo'
      auth_url: 'http://openstack/identity'

And then this can be referenced in the salt provider based on the democloud name.

  driver: openstack
  cloud: democloud
  region_name: RegionOne

This allows for just using one configuration for salt-cloud and for any other openstack tools which are all using /etc/openstack/clouds.yml

The other method allows for specifying everything in the provider config, instead of using the extra configuration file. This will allow for passing salt-cloud configs only through pillars for minions without having to write a clouds.yml file on each minion.abs

  driver: openstack
  region_name: RegionOne
    username: 'demo'
    password: secret
    project_name: 'demo'
    user_domain_name: default,
    project_domain_name: default,
    auth_url: 'http://openstack/identity'

Or if you need to use a profile to setup some extra stuff, it can be passed as a profile to use any of the vendor config options.

  driver: openstack
  profile: rackspace
    username: rackusername
    api_key: myapikey
  region_name: ORD
  auth_type: rackspace_apikey

And this will pull in the profile for rackspace and setup all the correct options for the auth_url and different api versions for services.


Most of the options for building servers are just passed on to the create_server function from shade.

The salt specific ones are:

  • ssh_key_file: The path to the ssh key that should be used to login to the machine to bootstrap it

  • ssh_key_file: The name of the keypair in openstack

  • userdata_template: The renderer to use if the userdata is a file that is templated. Default: False

  • ssh_interface: The interface to use to login for bootstrapping: public_ips, private_ips, floating_ips, fixed_ips

  • ignore_cidr: Specify a CIDR range of unreachable private addresses for salt to ignore when connecting

  provider: myopenstack
  image: CentOS 7
  size: ds1G
  ssh_key_name: mykey
  ssh_key_file: /root/.ssh/id_rsa

This is the minimum setup required.

If metadata is set to make sure that the host has finished setting up the wait_for_metadata can be set.

  provider: myopenstack
  image: CentOS 7
  size: ds1G
  ssh_key_name: mykey
  ssh_key_file: /root/.ssh/id_rsa
    build_config: rack_user_only
    rax_service_level_automation: Complete
    rackconnect_automation_status: DEPLOYED

If your OpenStack instances only have private IP addresses and a CIDR range of private addresses are not reachable from the salt-master, you may set your preference to have Salt ignore it:


Anything else from the create_server docs can be passed through here.

  • image: Image dict, name or ID to boot with. image is required unless boot_volume is given.

  • flavor: Flavor dict, name or ID to boot onto.

  • auto_ip: Whether to take actions to find a routable IP for the server. (defaults to True)

  • ips: List of IPs to attach to the server (defaults to None)

  • ip_pool: Name of the network or floating IP pool to get an address from. (defaults to None)

  • root_volume: Name or ID of a volume to boot from (defaults to None - deprecated, use boot_volume)

  • boot_volume: Name or ID of a volume to boot from (defaults to None)

  • terminate_volume: If booting from a volume, whether it should be deleted when the server is destroyed. (defaults to False)

  • volumes: (optional) A list of volumes to attach to the server

  • meta: (optional) A dict of arbitrary key/value metadata to store for this server. Both keys and values must be <=255 characters.

  • files: (optional, deprecated) A dict of files to overwrite on the server upon boot. Keys are file names (i.e. /etc/passwd) and values are the file contents (either as a string or as a file-like object). A maximum of five entries is allowed, and each file must be 10k or less.

  • reservation_id: a UUID for the set of servers being requested.

  • min_count: (optional extension) The minimum number of servers to launch.

  • max_count: (optional extension) The maximum number of servers to launch.

  • security_groups: A list of security group names

  • userdata: user data to pass to be exposed by the metadata server this can be a file type object as well or a string.

  • key_name: (optional extension) name of previously created keypair to inject into the instance.

  • availability_zone: Name of the availability zone for instance placement.

  • block_device_mapping: (optional) A list of dictionaries representing legacy block device mappings for this server. See documentation for details.

  • block_device_mapping_v2: (optional) A list of dictionaries representing block device mappings for this server. See v2 documentation for details.

  • nics: (optional extension) an ordered list of nics to be added to this server, with information about connected networks, fixed IPs, port etc.

  • scheduler_hints: (optional extension) arbitrary key-value pairs specified by the client to help boot an instance

  • config_drive: (optional extension) value for config drive either boolean, or volume-id

  • disk_config: (optional extension) control how the disk is partitioned when the server is created. possible values are 'AUTO' or 'MANUAL'.

  • admin_pass: (optional extension) add a user supplied admin password.

  • timeout: (optional) Seconds to wait, defaults to 60. See the wait parameter.

  • reuse_ips: (optional) Whether to attempt to reuse pre-existing floating ips should a floating IP be needed (defaults to True)

  • network: (optional) Network dict or name or ID to attach the server to. Mutually exclusive with the nics parameter. Can also be be a list of network names or IDs or network dicts.

  • boot_from_volume: Whether to boot from volume. 'boot_volume' implies True, but boot_from_volume=True with no boot_volume is valid and will create a volume from the image and use that.

  • volume_size: When booting an image from volume, how big should the created volume be? Defaults to 50.

  • nat_destination: Which network should a created floating IP be attached to, if it's not possible to infer from the cloud's configuration. (Optional, defaults to None)

  • group: ServerGroup dict, name or id to boot the server in. If a group is provided in both scheduler_hints and in the group param, the group param will win. (Optional, defaults to None)


If there is anything added, that is not in this list, it can be added to an extras dictionary for the profile, and that will be to the create_server function., call=None)

List available images for OpenStack

CLI Example

salt-cloud -f avail_images myopenstack
salt-cloud --list-images myopenstack, call=None)

List available sizes for OpenStack

CLI Example

salt-cloud -f avail_sizes myopenstack
salt-cloud --list-sizes myopenstack, call=None, kwargs=None)

Call function from shade.


function to call from shade.openstackcloud library

CLI Example

salt-cloud -f call myopenstack func=list_images
t sujksalt-cloud -f call myopenstack func=create_network name=mysubnet

Create a single VM from a data dict, conn=None, call=None)

Delete a single VM

Return the first configured instance.

Return a conn object for the passed VM data

Warn if dependencies aren't met., ip)

Return True if we are to ignore the specified IP., call=None)

List networks for OpenStack

CLI Example

salt-cloud -f list_networks myopenstack, call=None)

Return a list of VMs

CLI Example

salt-cloud -f list_nodes myopenstack, call=None)

Return a list of VMs with all the information about them

CLI Example

salt-cloud -f list_nodes_full myopenstack, call=None)

Return a list of VMs with minimal information

CLI Example

salt-cloud -f list_nodes_min myopenstack, call=None)

Return a list of VMs with the fields from query.selection

CLI Example

salt-cloud -f list_nodes_full myopenstack, call=None, kwargs=None)

List subnets in a virtual network


network to list subnets of

salt-cloud -f list_subnets myopenstack network=salt-net, ips)

Return either an 'ipv4' (default) or 'ipv6' address depending on 'protocol' option. The list of 'ipv4' IPs is filtered by ignore_cidr() to remove any unreachable private addresses., conn=None, call=None)

Request an instance to be built, conn=None, call=None)

Get VM on this OpenStack account


name of the instance

CLI Example

salt-cloud -a show_instance myserver

Return the ssh_interface type to connect to. Either 'public_ips' (default) or 'private_ips'.